The Mystery of Flight 427

by Bill Adair

  Brett listened intently. It sounded like the 737 was getting fixed.

  “Boeing has developed modifications to the rudders of older 737s that will improve safety,” Gore continued, “and they are going to begin retrofitting those planes, largely at their own expense, without waiting for a government mandate. Under a schedule to be developed by the FAA, these improvements will be made in the next two years. This is a major action: it affects some 2,800 planes worldwide, 1,100 of them here in the United States.”

  Tears welled up in Brett’s eyes. It sounded as if the government was actually doing something to prevent another crash. He felt vindicated. He had believed all along that there was some kind of flaw in the rudder system. He felt as though he had reached the end of a long road.

  At a news conference later that day, Boeing officials announced the specifics. The company would modify every 737 rudder valve so it could not reverse. A limiter would be installed on the PCU to prevent the rudder from going hardover. And, in a reminder that Boeing still believed the pilots were at fault, the company said it would pay for sensors on the rudder pedals. The next time a pilot stomped on the pedal by mistake, it would show up on the flight data recorder.

  Haueter was disappointed that Gore praised Boeing repeatedly and made only a passing mention of the NTSB. Gore made it sound as if Boeing, out of the kindness of its heart, had generously offered to spend $150 million to fix the PCUs. Never mind that the changes were a direct result of the work by him and Phillips!

  “We dragged these people kicking and screaming for the last two years, and all of a sudden they are getting all the credit,” Haueter said. “We have just been pounding them and pounding them, and now we don’t exist.”

  His boss, Bernard Loeb, shared his anger. “Boeing didn’t do this because their hearts told them to do it. Their lawyers told them to do it,” Loeb said. “They didn’t have a goddamn choice.”

  The same thing had happened two months earlier when Boeing made the dramatic discovery about the reversal. Whom did Boeing call then? The FAA. The company acted like the NTSB didn’t matter. Haueter had called Howes, the Boeing coordinator, and hollered at him about the fact that Boeing had not notified the NTSB about the thermal shock finding. Howes had sat through the entire Halloween meeting without saying a word. But he insisted to Haueter that no one at Boeing had told him about it.

  Haueter had discovered the downside of being the watchdog. To the public and the news media, NTSB investigators were crusaders for safety who could do no wrong. They were the guys in the white hats who found everybody else’s mistakes. But when given the chance, the groups that the NTSB had attacked welcomed the opportunity to fire a few zingers back. So when Boeing and the FAA announced a safety fix, they rarely gave the NTSB credit.

  The Boeing announcement was shrewdly timed. Haueter had been working on a new set of recommendations that called for immediate pilot training about the risk of a hardover. He felt that Boeing was trying to preempt him by announcing the rudder system changes first.

  Boeing also managed to steal the thunder from Dateline, the NBC newsmagazine that had been working on a segment about 737 rudders for several months. A Dateline crew had been out to Renton and interviewed several Boeing officials. Boeing made sure that Dateline correspondent Chris Hanson got a chance to ride in M-Cab. Two weeks before Gore’s announcement, rumors circulated through the investigators that Dateline was about to slam Boeing for the rudder problems. The segment was scheduled to air January 19.

  But then came Gore’s announcement on January 15, with Boeing saying it would improve the plane. That took the wind out of the Dateline segment, which ended up being a surprisingly positive piece that said the airplane was getting fixed.

  22. GRUNTS

  Malcolm Brenner, the eccentric NTSB psychologist, had made his name at the safety board by studying a drunken sailor. He and Jim Cash, the sound expert, had done a groundbreaking study of radio tapes of the Exxon Valdez accident that indicated the ship’s captain was drunk. To show he was intoxicated just before the ship ran aground, they counted the number of seconds it took him to say each word. He slurred the phrase “Exxon Valdez,” saying it 24 percent slower right before the accident than he had thirty-three hours earlier.

  Speech analysis was still a very new tool for plane crash investigations. For a 1985 Japan Air Lines crash, Japanese investigators measured the pitch of the pilots’ words to calculate when they were affected by stress after a sudden decompression. According to the investigators’ calculations, a calm Japanese male spoke at a frequency of 150 hertz, but they found that the pitch of the captain’s voice got as high as 410 hertz after the pilots heard a bang. The pitch of their voices also indicated the pilots were feeling the effects of hypoxia because they had neglected to don their oxygen masks.

  Brenner decided to use the same technique on Flight 427’s tape. The result might settle the debate over the Boeing theory that one of the pilots panicked because of the wake turbulence and slammed his foot on the rudder pedal. Brenner also hoped to look for correlations between the pilots’ grunts and when the rudder moved, which might indicate when they had pushed on the pedals.

  Brenner and three outside experts studied the cockpit tape one word at a time. Using Cash’s WAVES program, they counted syllables and measured the pitch and volume of each sentence. Germano had said “four-twenty-seven” seventeen times during the tape, which gave them a consistent phrase to measure. As stress increases, humans raise the pitch of their voices. Anything above 300 hertz is considered screaming, an indication that the person has panicked.

  They analyzed the way Germano said it each time. His average pitch was 144 hertz before the plane hit the wake turbulence, but it jumped to 214 hertz the last time he said it, which was about seventeen seconds after the wake turbulence. When they looked at his other comments on a bar graph, they saw a gradual increase in his frequency, which indicated that his stress level had progressively increased until he screamed and panicked at the end. There was no evidence that Germano had become “overaroused” when they hit the wake, as Boeing had suggested. They also found no evidence in Germano’s voice that he was trying to forcibly move the controls during the emergency. That was an important detail because it focused attention on Emmett, who had been the flying pilot on the Chicago-Pittsburgh leg of the trip.

  The results were less conclusive about Emmett. Scott Meyer, a navy sound expert, said Emmett’s grunts indicated that he was straining, but it was not clear whether the straining came from pushing his feet on the rudder pedals or moving the wheel with his arms. The sounds of straining stopped once the autopilot was switched off, which allowed Emmett to turn the wheel more easily.

  Alfred S. Belan, a sound expert from Russia who had analyzed hundreds of cockpit tapes, used a novel approach to figure out what Emmett did. He analyzed the word “shit.”

  Emmett had said the word during a calm moment about twenty minutes before the crash. He was trying to program the plane’s navigational computer, but kept having trouble. “Aw c’mon, you piece of shit!” he said. He said it two more times—once as the plane was starting to point nose down and a final time just before the crash. Belan compared his breathing each time and studied color graphs from Cash’s WAVES program. The graphs turned “shit” into orange lines that looked like a salmon filet, but Belan could look at them and see the difference between a grunt, Emmett’s inhaling, and the word.

  When people make a great physical effort, they take forced and rapid breaths. When Emmett said, “Aw c’mon, you piece of shit” to the flight-management computer, his breathing was normal. But after the plane was jostled by the wake, he grunted, and his voice showed signs of straining. When Emmett said the word seven seconds later, he said it softly and was not straining. That suggested he was no longer fighting the wheel or rudder pedals. It was easier to turn the wheel at that point because the autopilot had just been switched off. The final time Emmett said “Shit!” he was straining again, probably beca
use he was pulling back on the control column, trying to pull the plane’s nose up as the ground loomed closer.

  Brenner thought that the sound analysis disproved Boeing’s allegations that the pilots panicked early. The change in pitch was more gradual, proving that they didn’t panic until well after the wake turbulence. Germano said “427 emergency!” to air traffic controllers seven seconds after the stickshaker went off, which Brenner regarded as a rational, constructive action—to let controllers know the plane was in trouble.

  Brenner’s biggest finding involved the grunts.

  He put Emmett’s heavy breathing and grunts on a time line and noticed that they perfectly matched the theory of a rudder reversal. They occurred during the crucial three-and-a-half-second period when investigators believed the reversal occurred.

  Emmett’s soft grunt at the beginning came four-tenths of a second after the rudder began to reverse, when he would have felt the pedal pushing back against his right foot. He grunted louder about one second later, the time when the rudder pedals would have fully snapped back against his pushing. It was a nearly perfect match that indicated Emmett had been struggling against the rudder pedal but couldn’t stop it from reversing.

  It was the best evidence yet that the rudder had reversed.

  In January of 1997 Haueter had a difficult time sleeping. Many nights he found himself lying awake in the big four-poster bed that he shared with his wife, worried that another 737 would crash. The FAA had issued a slew of airworthiness directives that mandated changes to the 737 and how it was flown, including many that had been requested by the NTSB, but the agency had not acted on the NTSB recommendation that Haueter considered the most important—fixing the rudder’s power control unit.

  It had been two months since Vice President Gore had said Boeing would fix it, but Haueter had seen little action. Where was the urgency? Boeing and the FAA were taking a leisurely pace. The FAA—particularly McSweeny—often sounded like an apologist for Boeing. The FAA had mandated a weekly test for rudder jams, but the test was virtually meaningless. It would tell pilots if the PCU was jammed at that instant, but it didn’t tell whether a PCU would jam in the future. Haueter thought that pilots needed to be warned more clearly about the problems.

  It was time to light a fire under the FAA.

  If Haueter had been a politician, he would just have held a news conference and spoken his mind. But that was not how the NTSB operated. There were unwritten rules of engagement about how to blast the FAA. It was done discreetly, usually by letter to the FAA administrator, which was conveniently faxed to aviation reporters around the country.

  Haueter believed the 737 no longer met federal safety standards. He talked with colleagues about whether he should recommend that the entire 737 fleet be grounded but decided he didn’t have enough evidence for such a drastic step. The plane’s fatal accident rate was low compared with those of other planes—even if the rudder PCU did not meet federal standards. The odds of a jam and a reversal were still remote. Nevertheless, he was worried there might be another crash.

  In the letter, he used strong words. Haueter had grown more self-assured in the two and a half years since the crash. He’d been promoted to chief of major investigations, which meant he was now the head of all the big accidents, including the ValuJet crash in Miami and the granddaddy of them all, TWA 800. He was not shy about blasting the FAA and Boeing when he felt he had to. The ten-page letter, signed by Hall and endorsed by the other four board members, laid out Haueter’s case about how the soda can valve on the USAir plane was unique and more prone to reverse. The letter called on the FAA to alert pilots about the potential for a reversal and to speed up the process to fix all PCUs in the fleet. His words were surprisingly frank. He said the 737 was not as safe as other planes because the PCU was vulnerable to a reversal.

  To make sure that the message got prominent play, Haueter, Loeb, and public affairs director Peter Goelz held a background briefing with reporters from the New York Times, the Washington Post, and the TV networks. Within hours, the news was spread to millions of viewers.

  “Federal air safety experts are now satisfied they know what caused the mystery crash of a USAir Boeing 737,” Dan Rather said a few hours later on the CBS Evening News. “It was the rudder.” Without naming anyone at the safety board, CBS reporter Bob Orr said investigators “believe a malfunction in the Boeing 737’s rudder system rolled the plane into a fatal dive. And they suspect a similar failure caused the crash of a United Airlines 737 in Colorado Springs in 1991.”

  NBC anchor Tom Brokaw said, “There may be an answer tonight to questions that for years have perplexed investigators of two deadly plane crashes. That answer is certain to raise new fears about the world’s most widely used airliner, the Boeing 737.”

  Haueter watched the NBC report at home and felt tremendous relief. His worries about the plane had been told to the world. He wasn’t holding a secret anymore. Everyone knew what he knew.

  In the meantime, Boeing had gone into damage control mode. It faxed a statement to the press that said: “Boeing is and has been working with suppliers on an already aggressive schedule” to fix the rudder system. The changes “will serve to make a safe airplane even safer.”

  That had become the 737’s mantra, chanted by people at Boeing, at the airlines, even at the FAA. It was the perfect phrase to put a positive spin on the message. The 737 wasn’t dangerous. They were simply making it better.

  Despite heavy news coverage about the NTSB’s letter, there was barely a ripple of reaction from the public. The New York Times reported “a collective shrug of indifference” from passengers. “Travel agents, corporate travel managers and other industry officials said yesterday that most passengers demonstrated unshaken confidence in the overall safety of air travel,” the Times said, “and appear to feel that the Government would have grounded the nation’s fleet of 1,100 737’s if they were truly dangerous.”

  Boeing’s deft response minimized the repercussions for the company. Its stock price had not been affected by any of the 737 announcements. Wall Street didn’t seem to notice what the NTSB said. Airlines seemed satisfied that Boeing was fixing any problems that the plane had, and they continued to flood the company with orders for new 737s.

  Boeing wanted more tests that would push a 737 to its limit. A Boeing test pilot would fly over the Pacific Ocean and try rudder hardovers and other maneuvers to gather new data about the crossover point. Cox thought the tests were unnecessary and were being done to help Boeing defend itself from lawsuits. Worse, Cox was concerned that there might be an accident, which would lead to the immediate grounding of the entire 737 fleet. He joked that 737s would make good restaurants, which was a good thing because that’s all they would be useful for if a Boeing test pilot dropped one into the Pacific.

  But when investigators arrived in Seattle for the tests in the first week of June 1997, Boeing shocked everyone. For the first time, company engineers acknowledged that a jam and a reversal in the PCU could match the kinematic estimates for Flight 427. Until that point, Boeing had adamantly resisted any suggestion that a reversal caused the crash. The Boeing position had been firm: The pilots screwed up. But new estimates by Boeing’s engineers provided a surprisingly close match between Flight 427’s rudder and results from the jam/reversal tests. Boeing stopped far short of saying the plane was to blame, but just an acknowledgment that it was possible was a huge step. It was like Ronald Reagan saying nice things about communism.

  Boeing’s new position won Haueter’s admiration. “Holy mackerel, I’m shocked,” he said. “They have been fairly up front lately.”

  Cox also praised his rivals from Boeing for being so honest with data that appeared to indict their product. “The entire week was a night-and-day difference,” Cox said. “They were very forthright, there was not any of the partisanship.”

  The flight test was uneventful. Boeing pilots compiled new data about the crossover point and hardovers and did not crash in
the Pacific. But the most revealing event came when the plane returned to Boeing Field for a ground demonstration of a rudder reversal. The plane, which had just come off the assembly line for Southwest Airlines, was fitted with the special rig in its tail to jam the valve.

  Haueter had asked for the demonstration to show what a rudder hardover would have felt like to Emmett and Germano. It took two steps to cause the reversal. You pressed down with one foot to move the valve’s secondary slide into position, and then you had to stomp quickly with the other foot, which resulted in the second rudder pedal’s snapping back, as if it was trying to throw your foot off.

  It was a warm, sunny day in Seattle, once again proving that the city’s reputation for rain was a myth perpetuated by natives who didn’t want anyone else to move to their beautiful city. It was so bright that people standing on the pavement beneath the plane had to wear sunglasses. In the cockpit, Brenner sat down in the right seat beside Mike Carriker, the Boeing test pilot. The six-foot-three Brenner was the same height as Emmett, and he had to put the seat all the way back to be comfortable. He slowly pushed each rudder pedal as far as it would go, as if he were doing a routine check of the rudder before takeoff. He then jammed his foot on the left pedal, causing the reversal. The left pedal began coming up, pushing relentlessly against his foot, fighting him all the way until it reached the upper stop. He did the test a dozen more times and found that the rudder pedal overpowered him every time. He was amazed by the force. In a fight with a jammed PCU, a human could not win.

  Brenner had spent two and a half years trying to understand what had happened in 427’s cockpit. He knew the pilots so well that he knew their pant sizes. He knew their allergies, their marital histories, and what they ordered from room service. He had heard the cockpit tape hundreds of times, had spent hours in M-Cab and weeks poring over the flight data. But not until this moment, when the pedal came snapping back, did he truly understand what had happened. It was creepy. Brenner felt as though he was reliving those horrifying seconds at 6,000 feet. For a brief moment, he became Emmett. Oh, yeah, I see zuh Jetstream. He pushed on the rudder pedal to recover from the wake but felt it snap back. Oh shit. It was unrelenting. No matter how hard he pushed, the pedal Ohhh shiiiiiiittttt kept fighting back as the plane rolled to the left and then spun out of control God! toward the gravel road in Hopewell. Nooooo.